Tuesdays to Thursdays are my worst days also, and looks like they are also the less recreational friendly days I believe. Brazilian nights (GMT-3) give me double my winrate, without filtering any specific day of the week.

Thanks

Great job. I think you are ready for 100nl.

September 2023:



I spent the first half of the month on a vacation, so I didn't play much poker. My volume was a bit better in the 2nd half, but not really at the level I'd want it to be either.

Therefore, my goal for October is to start taking poker more seriously - like a job. I'll have a schedule for playing 3 hours/day and I'll need to have a really good excuse for deviating from it (hitting a downswing will no longer be an acceptable excuse).

Some hands:

When that green line went down how many flips did you lose in a single session? Lol.

it does seem like an insane run bad session lmao

Come on buddy, lets crush on october !

Lately, I've been considering the strategic use of the "All-in Cash Out" option in order to potentially get a bit of extra EV out of the preflop all-in spots.

My plan was to take advantage of the bunching effect and to cash out in spots where I get it in preflop with, for example, 66 vs AK, and to not cash out when I'm the one with high cards.

The idea is that high card hands (AK/AQ...) should long term run above EV in preflop AI spots, while low card hands (66,98s....) should run under the EV, as it's more likely that the other players folded low cards. However, it was difficult for me to mathematically predict exactly how significant the EV difference of this would be over a large sample / how much extra EV I could gain from doing this.

So, to find this out, I analyzed my database (2.7m+ hands) in the spots where I got it in preflop with high/low cards to observe the difference in the EV/Net Won. The results were very surprising:
(I also posted graphs for the top regs in my pool)

Getting it in with high cards:


Getting it in with low cards:


Apparently, over my sample, we've all been running significantly under EV with the high cards in preflop AI spots and way above EV with low cards!

This completely contradicts the whole concept of the Bunching effect. I'm not sure whether my sample is too small, or I'm just misunderstanding something?

It would be really helpful if someone else here, with a significant sample, would post their results. I'd appreciate it a lot.

All of them

A bit lazy to post the screenshot but, QQ+ and AK over 1.9mm hands I got 1766 instances where we went all in preflop. Running slightly under EV, 29,702.7 bbs won, 29,983.32 evbbs won. I don´t know how and if it is possible to filter villain's holecards in hem3, only hero's, so this result does not confirm or deny anyhing. And what you said makes sense obviously, it's probably a sample size issue.

You'd likely want to use HRC to estimate the hole cards left at the node you're getting it in.

I spent a few minutes thinking about this and could only come up with one harebrained explanation. Could it be because of the blockers in our hand? When we have 88, we're doubleblocking 8x folds. Villains will be heavily weighted away from holding (thus, folding) 8x combos pre. When we have AK, we're only singleblocking Ax folds. Most of the Ax and Kx is folding anyway.

I don't know if this makes sense.

What are the exact parameters you're using zkesic, and I'll replicate with zone alias tomorrow (several million hands).

My hypothesis is that most of the time you get it in all in with low cards when deep in a cash game, its a pocket pair. And pocket pairs often win when the hand of villain doesnt make a pair. In other words pocket pairs are made hands that dont rely that much on making trips so if I have 88 and someone folded an 8 in EP its not a tragedy, 88 doesnt need an 8 on the board to win.

On the other hand High cards are often non made hands AK-AQ,AJs and they need a pair. So if someone folded A2o in EP and Q3o in MP its a tragedy thats 2 out of 6 = 33% of the outs that u need to hit in order to win against a pocket pair.

Also u might underestinate how biased towards folding low hands the bunching effect is. People often fold offsuit high cards with low cards, so I expect a lot fo A,K,Q,J folds in early and middle position. Not as much as low cards of course, but the difference shouldnt be big enough to affect the aforementioned made/non-made hand effect

When equity is calculated, all of this is already taken into the account, so it shouldn't affect the results here.

Hero Position: BTN/SB/BB
First Raiser: BTN/SB/BB
All-in Street : Preflop
Cards:


(I chose those filters for cards so that the final graphs are roughly of the same size (7-8k hands for me))

It is...? Not sure if you're misunderstanding me or I'm missing something. I'm saying fewer 8s are folded than equity math would suggest because we block the 8x folds with our cards. That's really accounted for?

I'm not sure why the sample for me is significantly smaller.

High cards:



Low cards:



Filter:

Probably because I short stack a bunch and get it in way lighter

Thanks for the data. It seems that in your sample the bunching effect works as it's supposed to.

We'd probably need much larger samples for them to be sufficient in this research.

Don't know if these are useful in this case, but @Tombos21 has some interesting threads about the bunching effect.

Are flops random? (bunching effect)

The Bunching Effect

[GTO] The absurdity of high resolution solutions - bunching effect

This is not an intuitive calculation, to do without software aid. What we can say for sure is that AK should have slightly more outs to hit a pair than 87, bvb srp.

Is there any software that can give us weighted ranges? E.g we input smth like Gtow, or whatever ranges you use, and for HJ onwards it shows the actual probability that we should be dealt AA, 72o etc?

Considering coding one for myself here, too much free time and it should be easy anyway

HRC seems to be able to do this.

You can think about it like this:

Let's say we get it in preflop with red 88 vs AA. We have around 20% chance of winning.

Now, before dealing the flop, the dealer randomly removes 10 of the cards from the deck and throws them away. Do our odds of winning now change because "we block the 8x cards that he could've thrown away"? The answer is no, because we actually don't block any of our outs - we only block the red 88, while our outs are the black 88.

A similar thing happens when we're talking about the cards that our opponents folded. The odds of them folding our outs (black 88) haven't changed, as we don't actually block them.

Yeah, I think you're correct. I was thinking there was something about the weight of the folds (rather than removing random cards from the deck), but that seems fake. The only combos we directly interact with are the other pocket pair combos, which are not relevant here. Good explanation!

I think it's pretty simple. For a preflop all-in to happen, 2 or more people need to have strong hands (to build the pot with raises, reraises etc).

If you are one of 3 or more people who VPIP preflop, and you are holding 66, then you are going to overrealize your equity, because the other 2 (or more) players have ranges with mostly high cards that block each other. If you then end up all in with 66 (with one or more of these players), you are going to over-realize your equity.

If you are instead one of only 2 players who VPIP preflop, and you go all in with the other player, while holding 66, then you will underrealize because of bunching, as expected.

Keep in mind that I'm talking about the amount of players that VPIP in the hand preflop, not the amount that end up all in.

I'm guessing the commonality or magnitude of the multiway VPIP scenario outweighs the heads up VPIP scenario.

But since no one else came up with this, and I've never used hand2note, there's a good chance I'm missing something too XD

The big majority of All-in hands between BTN-BB are HU VPIP scenarios.